Current Projects

Network Meta-analysis Comparing Yoga, Resistance Training and Aerobic Exercise on Heart Rate Variability
We searched published journal articles from PubMed, Embase and Scopus about Yoga and other non-pharmacological therapies impacting RSMMD. After screening and coding, I ranked the effectiveness of those therapies and helped resolve several controversial findings.

A Bayesian Framework for Correlated Continuous Outcomes Using Individual and Aggregate Data
We propose a Bayesian network meta-analysis (NMA) framework for jointly modeling correlated continuous outcomes, incorporating both individual participant data (IPD) and aggregate data (AD). This method addresses common challenges in clinical and epidemiological research, such as partial IPD availability and outcome correlation. Under the Bayesian framework, our approach borrows strength across treatments and outcomes, and models feature-specific effects using information from both data levels. Implemented in RJAGS, our method enables robust evidence synthesis for complex mixed-data problems. Simulations are used to evaluate bias and precision compared to standard NMA approaches.

The Association between Bronchial Hyperresponsiveness and Respiratory Traits Over Time
We assess the association between bronchial hyperresponsiveness (BHR) and respiratory traits such as asthma, lung function, airway inflammation, and remodeling over time. The statistical modeling is based on generalized linear regression with repeated measures.

Sample Selection from Big Data by Trimmed Space-Filling Based on PCA (Master’s Degree Project)
This project applies machine learning to analyze large-scale datasets, aiming to uncover hidden patterns and relationships. The trimmed space-filling sampling method selects a subset of data points by applying a space-filling strategy to the principal components extracted by PCA.